高可用flume-ng搭建

一、概述
1.通过搭建高可用flume来实现对数据的收集并存储到hdfs上，架构图如下：

二、配置Agent
1.cat flume-client.properties
#name the components on this agent  声明source、channel、sink的名称  
a1.sources = r1  
a1.sinks = k1 k2  
a1.channels = c1  
    
#Describe/configure the source    声明source的类型为通过tcp的方式监听本地端口5140  
a1.sources.r1.type = syslogtcp  
a1.sources.r1.port = 5140  
a1.sources.r1.host = localhost  
a1.sources.r1.channels = c1  
    
#define sinkgroups  此处配置k1、k2的组策略，类型为均衡负载方式  
a1.sinkgroups=g1  
a1.sinkgroups.g1.sinks=k1 k2  
a1.sinkgroups.g1.processor.type=load_balance  
a1.sinkgroups.g1.processor.backoff=true  
a1.sinkgroups.g1.processor.selector=round_robin  
    
#define the sink 1  数据流向，都是通过avro方式发到两台collector机器  
a1.sinks.k1.type=avro  
a1.sinks.k1.hostname=Hadoop1  
a1.sinks.k1.port=5150  
    
#define the sink 2  
a1.sinks.k2.type=avro  
a1.sinks.k2.hostname=hadoop2 
a1.sinks.k2.port=5150  
    
    
# Use a channel which buffers events in memory  指定channel的类型为内存模式 
a1.channels.c1.type = memory  
a1.channels.c1.capacity = 1000  
a1.channels.c1.transactionCapacity = 100  
    
# Bind the source and sink to the channel  
a1.sources.r1.channels = c1  
a1.sinks.k1.channel = c1  
a1.sinks.k2.channel=c1

#a2和a3的配置和a1相同

三、配置Collector
1.cat flume-server.properties
#name the components on this agent  声明source、channel、sink的名称 
collector1.sources = r1  
collector1.channels = c1 
collector1.sinks = k1  
    
# Describe the source  声明source的类型为avro 
collector1.sources.r1.type = avro  
collector1.sources.r1.port = 5150  
collector1.sources.r1.bind = 0.0.0.0  
collector1.sources.r1.channels = c1  
    
    
# Describe channels c1 which buffers events in memory 指定channel的类型为内存模式 
collector1.channels.c1.type = memory  
collector1.channels.c1.capacity = 1000  
collector1.channels.c1.transactionCapacity = 100  
    
# Describe the sink k1 to hdfs  指定sink数据流向hdfs 
collector1.sinks.k1.type = hdfs  
collector1.sinks.k1.channel = c1  
collector1.sinks.k1.hdfs.path = hdfs://master/user/flume/log
collector1.sinks.k1.hdfs.fileType = DataStream  
collector1.sinks.k1.hdfs.writeFormat = TEXT  
collector1.sinks.k1.hdfs.rollInterval = 300  
collector1.sinks.k1.hdfs.filePrefix = %Y-%m-%d  
collector1.sinks.k1.hdfs.round = true  
collector1.sinks.k1.hdfs.roundValue = 5  
collector1.sinks.k1.hdfs.roundUnit = minute  
collector1.sinks.k1.hdfs.useLocalTimeStamp = true

#collector2配置和collector1相同

四、启动
1.在Collector上启动fulme-ng

flume-ng agent -n collector1 -c conf -f /usr/local/flume/conf/flume-server.properties -Dflume.root.logger=INFO,console 
# -n 后面接配置文件中的Agent Name

2.在Agent上启动flume-ng

flume-ng agent -n a1 -c conf -f /usr/local/flume/conf/flume-client.properties -Dflume.root.logger=INFO,console

五、测试
[[email protected] ~]#  echo "hello" | nc localhost 5140    #需要安装nc

17/09/03 22:56:58 INFO source.AvroSource: Avro source r1 started. 
17/09/03 22:59:09 INFO ipc.NettyServer: [id: 0x60551752, /192.168.100.15:34310 => /192.168.100.11:5150] OPEN 
17/09/03 22:59:09 INFO ipc.NettyServer: [id: 0x60551752, /192.168.100.15:34310 => /192.168.100.11:5150] BOUND: /192.168.100.11:5150 
17/09/03 22:59:09 INFO ipc.NettyServer: [id: 0x60551752, /192.168.100.15:34310 => /192.168.100.11:5150] CONNECTED: /192.168.100.15:34310 
17/09/03 23:03:54 INFO hdfs.HDFSDataStream: Serializer = TEXT, UseRawLocalFileSystem = false
17/09/03 23:03:54 INFO hdfs.BucketWriter: Creating hdfs://master/user/flume/log/2017-09-03.1504494234038.tmp

六、总结
高可用flume-ng一般有两种模式：load_balance和failover。此次使用的是load_balance，failover的配置如下：
#set failover 
a1.sinkgroups.g1.processor.type = failover 
a1.sinkgroups.g1.processor.priority.k1 = 10 
a1.sinkgroups.g1.processor.priority.k2 = 1 
a1.sinkgroups.g1.processor.maxpenalty = 10000

一些常用的source、channel、sink类型如下：

本文永久更新链接地址：http://www.linuxidc.com/Linux/2017-10/147645.htm